To reproduce: 1. loggingmanager.register_sink_type('foo', function() error'boom' end) 2. reload logging 3. observe available log_sink_types When reloading the logging config all sinks are removed and stored in a temporary table, then they are added back, which triggers their application on the logging config. If an error interrupts this, any sinks that are still in the temporary table are lost. This can be fixed by reloading loggingmanager, like this in the telnet console: >require'core.loggingmanager'.reload_logging = function()end | Result: nil >package.loaded['core.loggingmanager']=nil | Result: nil >require'util.startup'.init_logging() | Result: nil

• title Permanent of log sinks in case of error during logging reload Permanent loss of log sinks in case of error during logging reload